Did you guys share the pretraining code

[menglin0320]

trafficstars

I'm reading the paper. And for our project, the pretraining part is more helpful so I wonder if you guys also shared these code somewhere? Also I have a question for the Detection of Direction Exceptions task. Why did you guys chose to use the majority direction in the positive set? Is it related to the way you guys sample the directions? If it's possible can I get more details?

[wdp-007]

Which paper are you reading?

[menglin0320]

[CVPR 2023] GeoLayoutLM: Geometric Pre-training for Visual Information Extraction

[ccx1997]

Why did you guys chose to use the majority direction in the positive set?

We assume that most relation predictions are right by a simple/coarse relation head in a real downstream task, which is reflected in DDE where the majority direction is regarded as positive.